Bayesian Prediction of Anxiety Level in Aged People at Rest Using 2-Channel NIRS Data from Prefrontal Cortex
The aim of this study was to predict mental stress levels of aged people at rest from two-channel near-infrared spectroscopy (NIRS) data from the prefrontal cortex (PFC). We used the State-Trait Anxiety Inventory (STAI) for the mental stress index.
We previously constructed a machine learning algorithm to predict mental stress level using two-channel NIRS data from the PFC in 19 subjects aged 20–24 years at rest (Sato et al., Adv Exp Med Biol 765:251–256, 2013). In the present study, we attempted the same prediction for aged subjects aged 61–79 years (10 women; 7 men). The mental stress index was again STAI. After subjects answered the STAI questionnaire, the NIRS device measured oxy- and deoxy-hemoglobin concentration changes during a 3-min resting state. The algorithm was formulated within a Bayesian machine learning framework and implemented by Markov Chain Monte Carlo. Leave-one-subject-out cross-validation was performed.
Average prediction error between the actual and predicted STAI values was 5.27. Prediction errors of 12 subjects were lower than 5.0. Since the STAI score ranged from 20 to 80, the algorithm appeared functional for aged subjects also.
KeywordsAnxiety Near infrared spectroscopy Prefrontal cortex Bayesian regression Aging
- 4.Sato M, Ishikawa W, Suzuki T, Matsumoto T, Tsujii T, Sakatani K (2013) Bayesian STAI anxiety index predictions based on prefrontal cortex NIRS data for the resting state. Adv Exp Med Biol 765:251–256. doi:10.1007/978-1-4614- 4989-8_35Google Scholar
- 5.Bishop C (2006) Pattern recognition and machine learning. Springer, HeidelbergGoogle Scholar